Overview

In Experiment 1, we show that children make a so-called mutual exclusivity inference and that this inference depends on children’s developing semantic knowledge. In Experiment 2, we show that children make inferences about word meanings based on common ground. In Experiment 3, we show that, when combined in one procedure, children are sensitive to the way that the two inferences are aligned.

Methods

Participants

A total of 41 great apes participated at least once in one of the tasks. This included 8 Bonobos (3 females, age 7.3 to 38.2), 22 Chimpanzees (17 females, age 2.6 to 55.2), 6 Gorillas (4 females, age 2.7 to 21.9), and 6 Orangutans (4 females, age 17 to 40.5). The sample size at the different time points ranged from 0 to 21. Figure visualizes the sample size across time points. We tried to test all apes at all time points but this was not always possible due to a lack of motivation or construction works. All apes participate in cognitive research on a regular basis. Many of them have ample experience with the very tasks we used in the current study.

Apes were housed at the Wolfgang Köhler Primate Research Center located in Zoo Leipzig, Germany. They lived in groups, with one group per species and two chimpanzee groups. Research was noninvasive and strictly adhered to the legal requirements in Germany. Animal husbandry and research complied with the European Association of Zoos and Aquaria Minimum Standards for the Accommodation and Care of Animals in Zoos and Aquaria as well as the World Association of Zoos and Aquariums Ethical Guidelines for the Conduct of Research on Animals by Zoos and Aquariums. Participation was voluntary, all food was given in addition to the daily diet, and water was available ad libitum throughout the study. The study was approved by an internal ethics committee at the Max Planck Institute for Evolutionary Anthropology.

Sample size by species across the different time points. Time point specific predictor variables were collected during the time between two time points (shaded regions) to predict the next.

Figure 1: Sample size by species across the different time points. Time point specific predictor variables were collected during the time between two time points (shaded regions) to predict the next.

Tasks

Apes were tested in familiar sleeping or observation rooms by a single experimenter. Whenever possible, they were tested individually. The basic setup comprised a sliding table positioned in front of a clear Plexiglas panel with three holes in it. The experimenter sat on a small stool and used an occluder to cover the sliding table.

Setup used for the six tasks. A) Causal reasoning and inference by exclusion. B) Quantity discrimination. C) Gaze following. D) Switching. E) Delay of gratification.

Figure 2: Setup used for the six tasks. A) Causal reasoning and inference by exclusion. B) Quantity discrimination. C) Gaze following. D) Switching. E) Delay of gratification.

Gaze Following

The gaze following task was modeled after Brauer, Call, and Tomasello (2005). The experimenter sat opposite the ape and handed over food at a constant pace. That is, the experimenter picked up a piece of food, briefly held it out in front of her face and then handed it over to the participant. After a predetermined (but varying) number of food items had been handed over, the experimenter again picked up a food item, held it in front of her face and then looked up (i.e., moving her head up). After 10s, the experimenter looked down again and handed over the food. We coded whether the subject looked up during the 10s interval. Participants received a total of 8 trials, spread out across the two test days. Each session started with two

Causal inference

The causality and inference tasks were modeled after (Call 2004). Two identical cups with a lid were placed left and right on the table. The experimenter covered the table with the occluder, retrieved a piece of food, showed it to the ape, and hid it in one the cups outside the participant’s view. Next, they removed the occluder, picked up the baited cup and shook it three times, which produced a rattling sound. Next, the cup was put back in place, the sliding table pushed forwards, and the participant made a choice by pointing to one of the cups. If they picked the baited cup, their choice was coded as correct, and they received the reward. If they chose the other cup, they did not. On each time point, participants received 12 trials.

Inference by exclusion

Inference trials were identical to causality trials, but instead of shaking the baited cup, the experimenter shook the empty cup. Correct choice was again coded when the baited (non-shaken) cup was chosen. On each time point, participants received 12 trials. Inference trials were intermixed with causality trials.

Quantity discrimination

For this task, we followed the general procedure of (Hanus and Call 2007). Two small plates were presented left and right on the table. The experimenter placed 5 small food pieces on one plate and 7 on the other. Then they pushed the sliding table forwards, and the subject made a choice. We coded as correct when the subject chose the plate with the larger quantity. There were 12 trials per time point.

Switching

This task was modeled after (Haun et al. 2006). Three differently looking cups (metal cup with handle, red plastic ice cone, red cup without handle) were placed next to each other on the table. There were two conditions. In the place condition, the experimenter hid a piece of food under one of the cups in full view of the participant. Next, the cups were covered by the occluder and the experimenter switched the position of two cups, while the reward remained in the same location. We coded as correct if the participant chose the location where the food was hidden. Participants received four trials in this condition. The place condition was run first. The feature condition followed the same procedure, but now the experimenter also moved the reward when switching the cups. The switch between conditions happened without informing the participant in any way. A correct choice in this condition meant choosing the location to which the cup plus the food were moved. Here, participants received eight trials. The dependent measure of interest for this task was calculated as: [proportion correct place] - (1 - [proportion correct feature]). Positive values in this score mean that participants could quickly switch from choosing based on location to choosing based on feature. High negative values suggest that participants did not or hardly switch strategies.

Delay of gratification

Data collection

sessions For each individual, the tasks at one time point were usually spread out across two consecutive days with causality and inference on day 1 and quantity and switching on day 2. Gaze following trials were run at the beginning and the end of each day.

phases

Predictors

Stable individual characterisitcs

Variable individual characteristics

Group life

Testing arrangements

Analytical framework

Structural equation modelling

Stability, reliability, relations between tasks

Simulations

Projection predictive inference

predictor selection

Results

Results from the five cognitive tasks across time points. Black crosses show mean performance at each time point across species (with 95\% CI). Colored dots show mean performance by species. Transparent grey lines connect individual performances across time points, with the line’s width corresponding to the number of participants. Dashed line shows the chance level inference whenever applicable. The panel for switching includes triangles and dots showing the mean performance in the two phases from which the overall performance score was computed.

Figure 3: Results from the five cognitive tasks across time points. Black crosses show mean performance at each time point across species (with 95% CI). Colored dots show mean performance by species. Transparent grey lines connect individual performances across time points, with the line’s width corresponding to the number of participants. Dashed line shows the chance level inference whenever applicable. The panel for switching includes triangles and dots showing the mean performance in the two phases from which the overall performance score was computed.

Phase 1

Stability

Reliability

Relations between tasks

Summary

Appendix

Brauer, Juliane, Josep Call, and Michael Tomasello. 2005. “All Great Ape Species Follow Gaze to Distant Locations and Around Barriers.” Journal of Comparative Psychology 119 (2): 145.

Call, Josep. 2004. “Inferences About the Location of Food in the Great Apes (Pan Paniscus, Pan Troglodytes, Gorilla Gorilla, and Pongo Pygmaeus).” Journal of Comparative Psychology 118 (2): 232.

Hanus, Daniel, and Josep Call. 2007. “Discrete Quantity Judgments in the Great Apes (Pan Paniscus, Pan Troglodytes, Gorilla Gorilla, Pongo Pygmaeus): The Effect of Presenting Whole Sets Versus Item-by-Item.” Journal of Comparative Psychology 121 (3): 241.

Haun, Daniel BM, Josep Call, Gabriele Janzen, and Stephen C Levinson. 2006. “Evolutionary Psychology of Spatial Representations in the Hominidae.” Current Biology 16 (17): 1736–40.